Analytical Methods For Effective TCM Clinical Formula Knowledge Discovery

Traditional Chinese Medicine (TCM) with its particular theoretical system and diagnostic method has been gradually developed from long-term clinical practices. TCM diagnostic and treatment, as the typical "individualized diagnosis and treatment" method, start with the observation and identification of individual clinical manifestations (mainly including symptom and sign). Prescription, the major form of TCM clinical medicine, is not only the important means of complex intervention during the medical practice, but the main knowledge sources of new significant herb detection. With the accumulation of large scale clinical data, using the data mining method,the effective core formula discovery based on the certain disease has become the efficient pathway and hot orientation of new significant herb detection.Our study starts from the relationship analysis between two most basic medical entities (symptom and herm) in TCM clinical diagnosis and treatment. Through using the efficacy evaluation information, it explored the analytical method of clinical effective formula detection based on the large scale TCM diagnosis and treatment database in the real world. Integrating the inside and outside relationship of herb, symptom, formula and curative effect of the individual patient in TCM, we research the method of herb-symptom discovery and effective formula detection. The major research work including:(1)Herb syndrome correspondence phenomenon and herb-symptom knowledge analysis method. This method carried out related modeling and study on representative dataset about four high quality dataset (the patient cases with liver-spleen disharmony, referred to as GPBT; insomnia medical cases, referred to as INSOMNIA; the clinical cases of children with Tourette’s syndrome, referred to as TS; the inpatient cases with congestive heart failure, referred to as CHF) from the real word. It proceeds from the holistic verification on basic method and principal Herb syndrome correspondence (Pearson coefficient is0.96), and proposals a network-based correlation analysis method (NetCorrA) based on Chi-Square Test and herb efficacy. After clinical evaluation to excavate relation knowledge between herb and symptom which hiding behind the TCM clinical data. After clinical evaluation, we see that NetCorrA can effectively remove the false positive correlation results which exist in the Chi-Square Test, and show the great analytical effect.(2)The effective formula and symptomatic knowledge analysis method, which utilizes herb set enrichment analysis method (HSEA) to enrich the effective formula and conditional mutual information method (CMI) and global silencing of indirect correlations method (GSIC) to detect symptomatic knowledge. To Balance the confounding factors of the samples and eliminate cofounding bias, we apply propensity score matching method on the dataset of disease insomnia which comes from TCM clinical to get the180samples for experiment. Hierarchical core structure extraction method and herb set enrichment analysis (HSEA) were adopted to explore15effective clinical formulae whose P-value<0.05. The direct correlation symptomatic knowledge, of which CMIS are above0.70, was found by conditional mutual information and global silencing of indirect correlations method (GSIC). GSIC can effectively eliminate the indirect interacting correlations between herb and symptom from the perspective of complex networks. Finally, the analysis method detecting effect formula and individual herb compatibility based on clinical data were formed.